How many scientists does it take to change a lightbulb?

More on the lightbulb situation.

The Australians are likely to be followed by a number of other countries, for the simple fact that incandescent lights are so inefficient, and lighting is so widely used. The effect of this is huge: Global CO2 emissions from lighting were 70% of that from all the planets cars (International Energy Agency report, 2006)

95% of the energy from a tungsten light bulb goes in heat - and that is largely heating just the ceiling. And these make up 79% of the global lighting sales.

Compact fluorescent lightbulbs are better. Slightly - they're only 85% inefficient.

So what's the alternative? One is solid-state lighting. The US Department of Energy calculated that if US households were able to use GaN-based LEDs instead of incandescent lighting, they could close 41 power stations.

So what's stopping them? Well, to be commercially viable, GaN-based white LEDs have to be high efficiency, high quality, long life and low cost. At the moment, commercial LEDs get up to around 10% efficient, not even beating fluorescent lamps. Lab produced LEDs have managed to match high-pressure Sodium lamps for efficiency.

Lifetime and reliability is a problem - I've heard rumours that the Chinese had decided to use large numbers of white LEDs for lighting for infrastructure connected to the 2008 Olympic Games, and they are already having to replace large numbers due to failure. Manufacturers claim 100,000 hours lifetime, but tests by Cambridge University have shown this isn't true - the problem often isn't the LED itself, but the wire bonds. Basically, you get what you pay for - expensively produced, well sealed white LEDs (e.g. in silicone) do last a very long time. Cheaper epoxies put more stress on the wire bond, and fail quicker.

So, what's the cost? For widespread adoption of white LEDs, you need about $5 cost per 1000 lumens. The current cost of quality white LEDs? Around $100.

There's a long way to go yet.